Research Article
Online Missing Data Imputation Using Virtual Temporal Neighbor in Wireless Sensor Networks
Algorithm 2
VTN imputation algorithm.
Input: input data matrix (), node , sensor , , | | Output: input data matrix () | | if there exists a new data in IDM then | (27) | | (28) | U | (29) | V | (30) | vtn | (31) | Get the new data from IDM by v_index | (32) | Call Calculation of VTN | (33) | | (34) | if then | (35) | Add c_index into imputation_index | (36) | end if | (37) | if then | (38) | for each element in do | (39) | if then | (40) | Add into imputation_index | (41) | end if | (42) | end for | (43) | end if | (44) | if then | (45) | Sort imputation_index by increasing index | (46) | for each element j in imputation_index do | (47) | for each k in 1: j do | (48) | if then | (49) | Add into | (50) | end if | (51) | end for | (52) | sort by increasing order of | (53) | | (54) | | (55) | high | (56) | if then | (57) | | (58) | else if high | (59) | high | (60) | end if | (61) | | (62) | | (63) | for each element m in PAST_VTNM_CANDADATE do | (64) | CHANGE_RATE_DIST | (65) | | (66) | end for | (67) | sort by increasing order of CHANGE_RATE_DIST | (68) | Remove imputing value from | (69) | | (70) | | (71) | Construct the estimation equation using PAST_VTNM_CANDADATE and IDM_CANDIDATE to | (72) | regress the coefficients | (73) | Compute using and | (74) | | (75) | end for | (76) | end if | (77) | end if | (78) | return IDM |
|